Journal of Archaeological Science (2003) 30, 439–459doi:10.1006/jasc.2002.0853

Geo-Ethnoarchaeology of Pastoral Sites: The Identification ofLivestock Enclosures in Abandoned Maasai Settlements

Ruth Shahack-Gross

Anthropology Dept., Washington University, One Brookings Drive, St Louis, MO 63130, U.S.A. andDept. of Structural Biology, Weizmann Institute of Science, Rehovot 76100, Israel

Fiona Marshall

Anthropology Dept., Washington University, One Brookings Drive, St Louis, MO 63130, U.S.A.

Steve Weiner*

Dept. of Structural Biology, Weizmann Institute of Science, Rehovot 76100, Israel

(Received 21 December 2001, revised manuscript accepted 14 May 2002)

The earliest food producers in Africa were mobile pastoralists who left limited archaeological traces. As a resultarchaeologists studying the spread of food production in the region have difficulty distinguishing early pastoralists fromhunter-gatherers with whom they interacted. This geo-ethnoarchaeological study contributes to the resolution of theproblem through identification of sediments distinctive of livestock enclosures, and thus of pastoral settlements.Sediments were sampled in and around currently occupied and recently abandoned Maasai livestock enclosures rangingin age between one and 40 years. Twenty to thirty years after site abandonment, there is no visible difference betweenenclosure and regional sediments. Micromorphological, mineralogical, and phytolith analyses, of enclosure sediments,however, allow differentiation of enclosure from regional sediments. Our results show that a unique undulatingmicrolaminated structure is distinctive of enclosure sediments. Enclosure sediments, especially small stock, also containa rare mineral, monohydrocalcite (CaCO3 . H2O). In addition, large amounts of opal (SiO2 . nH2O), in the form ofphytoliths, are found in enclosure relative to regional sediments. These differences are likely to be preserved in thearchaeological record, and this approach will allow better understanding of the spread of pastoralism in Africa andelsewhere. � 2002 Elsevier Science Ltd. All rights reserved.

Keywords: PASTORALISM; EAST AFRICA; ETHNOARCHAEOLOGY; MINERALOGY;MICROMORPHOLOGY; PHYTOLITHS; LIVESTOCK ENCLOSURES.

Introduction

T he adoption of food production, and a newmore intensive relationship with plant andanimal resources, represents a major turning

point in prehistory (reviews in Gebauer & Price, 1992;Price & Gebauer, 1995; Harris, 1996). The question ofwhy and how this happened is one of the centralquestions in archaeology. In Africa, pastoralism is theearliest form of food production, preceding mixedagriculture by 4000 years (van der Veen, 1999; Blench& MacDonald, 2000; Marshall & Hildebrand, 2002 ).Pastoralists, mobile herders who move animals towater and pasture, are also known from many otherareas of the world (Dyson-Hudson & Dyson-Hudson,

4390305–4403/03/$-see front matter

1980). In order to understand the earliest food produc-tion in Africa, and the spread of pastoralism elsewhere,it is important to be able to identify archaeologicalsites that were occupied by pastoralists, and distinguishthem from those occupied by hunter-gatherers.

The archaeological problemThe identification of archaeological sites occupied bypastoralists is not always straight-forward. First, basedon ethnographic observations, pastoralists tend toleave few material remains in their abandoned sites(e.g., Gifford, 1978; Robertshaw, 1978; Banning& Kohler-Rollefson, 1992). Second, many pastoralsites preserve only a few features (Robertshaw &Marshall, 1990; Mutundu, 1999). Third, interactions

*Corresponding author.

� 2002 Elsevier Science Ltd. All rights reserved.

440 R. Shahack-Gross et al.

between pastoralists and hunter-gatherers or horticul-turalists may result in ambiguous material remains inarchaeological sites (elaborated below). The difficultyof distinguishing between hunter-gatherer and pastoralsites, in particular, causes a major interpretive problemin African prehistory.

In East Africa, prehistoric hunter-gatherer andpastoral groups (4000–2000 ) have been most com-monly differentiated based on distinctive lithic assem-blages (Ambrose, 1984, 2001). Ceramics of the period,are believed to have been mostly produced bypastoralists (Ambrose, 1984; Gifford-Gonzalez, 1998).Interactions between hunter-gatherers and pastoralistsare recognized archaeologically by the occurrence ofmixed faunal assemblages (i.e. remains of wild anddomestic taxa at the same site, as at Enkapune YaMuto; Ambrose, 1998), and the presence of pastoralceramics in hunter-gatherer sites (Ambrose, 1998;Gifford-Gonzalez, 1998). Due to these interactions, itis sometimes difficult to distinguish between sites thatwere occupied by hunter-gatherers and those occupiedby pastoralists. In addition, where hunter-gatherer sitespreserve domestic stock, it is difficult to know whetherhunter-gatherers acquired these animals for immediateconsumption or whether they are in the process ofbecoming herders (Marean, 1992). These problems arecentral to understanding the spread of food produc-tion, particularly in eastern and southern Africa(Bower, 1991; Marshall, 1990; Smith, 1992).

Pastoral groups live in open-air settlements thatinclude fenced animal enclosures in order to protecttheir herds from nocturnal predators and from raidingby other humans (Kruuk, 1980; Marshall, 2000). Incontrast, hunter-gatherer settlements do not includeanimal enclosures, unless the hunter-gatherers are inthe process of adoption of pastoralism (Yellen, 1984;Mutundu, 1999). One strategy for identifying pastoralsites is to take advantage of the fact that in East Africa(and elsewhere) large amounts of dung accumulate inlivestock enclosures (e.g., Gifford, 1978; Mbae, 1986;Lamprey & Waller, 1990). In open-air archaeologicalsites, however, dung, which is mostly composed oforganic matter, degrades with time and is not readilyidentified. The major objective of this study, therefore,is to find durable indicators for livestock enclosures.

Previous researchPrevious research dealing with identification of live-stock dung is mainly limited to relatively rare depositsthat preserve organic remains. Such studies includethose of lipid biomarkers (e.g., Evershed et al., 1997;Simpson et al., 1998; Bull et al., 1999), parasites indung (e.g., Jones et al., 1988; Schelvis, 1992), andmicromorphology. Micromorphological studies mainlybase the identification of dung deposits on the presenceof organic fibres and structureless (amorphous) organicmatter. Mineralogical components associated withthe organic matter are also used. These include opal

phytoliths, calcium-oxalate druses, phosphate com-pounds, and in particular dung spherulites (Courtyet al., 1991; Macphail & Goldberg, 1995; di Lernia,1998; Boschian & Montagnari-Kokelj, 2000; Akeret &Rentzel, 2001; see also Brochier et al., 1992). Wattezet al. (1990) identified a microlaminated structure insediments from the cave of Arene Candide that isbelieved to be derived from herbivorous excrements.

Brochier (1983) emphasized that the association ofspherulites and phytoliths is important for identify-ing ancient pastoral activities. Dung spherulites arecalcareous spheres measuring 5–15 �m in diameter,that possess a spherulitic figure (i.e. a pseudo-uniaxialinterference figure) when observed under crossedpolarized light (Canti, 1997, 1998). Spherulites form inthe guts of animals (Canti, 1999) and are excretedin their dung. Caprine dung is particularly rich inspherulites (Canti, 1998). Spherulites do, however,occur in the dung of other animal species, not allof them domestic (Canti, 1997, 1998; Goren, 1999).Brochier et al. (1992) note that spherulites do notpreserve in open-air sites.

Phytolith morphologies have been used to distin-guish cattle from sheep dung (Powers et al., 1989).Elemental analyses (mainly phosphorous) have beenused to identify activity areas within sites, but are notindicative of specific activities (e.g., Lillios, 1992;Sanchez et al., 1996; see also Middleton & Price, 1996).

In this study we searched for durable indicators oflivestock enclosure sediments by using a combinationof techniques including micromorphology, mineralogyand quantitative phytolith analyses. All analyseswere performed following geo-ethnoarchaeologicalfieldwork in southern Kenya.

StrategyFieldwork was conducted among the Kisongo Maasaiof Kajiado District, near Rombo (Figure 1). TheKisongo Maasai are subsistence-oriented cattle, sheepand goat pastoralists. Their mode of pastoralism issemi-nomadic, taking advantage of seasonal pastureareas (Ryan et al., 2000; on Maasai pastoralism ingeneral see Dyson-Hudson & Dyson-Hudson, 1980;Galaty, 1981; Grandin et al., 1989; Lamprey & Waller,1990). A Maasai settlement, called boma (Swahili) orenkang (Maa), is made up of several independentpolygamous families and their livestock. Settlementsare laid out around a central livestock enclosureformed by encircling huts and thorn fences, andsmaller enclosures for sheep and goats (i.e. caprines)and calves (Mbae, 1986; pers. obs., 1999; Figure 2).The huts are constructed from a mixture of ash, cattledung and mud over a wooden frame (Andersen, 1978;pers. obs., 1999). Dung is never burned for fuel.Maasai settlements in general, and caprine enclosuresin particular, contain thick dung deposits (pers. obs.,1999). Because contemporary Maasai pastoralists liveunder similar environmental conditions as prehistoric

Geo-Ethnoarchaeology of Pastoral Sites 441

East African pastoralists, and herd the same domesticspecies, we expect that patterns of dung buildup inMaasai settlements are analogous to dung buildupsin prehistoric pastoral sites.

Two seasons of fieldwork (May 1999, January 2001)oriented particularly towards obtaining sedimentsamples for further study, were conducted by one of us(R. S-G.) in collaboration with K. Ryan (MASCA,The University Museum of Archaeology and Anthro-pology, Philadelphia) and Dr Karega-Munene(National Museums of Kenya, Nairobi). Sedimentsfrom one occupied boma and four abandoned bomaswere sampled, and compared to regional sedimentsoutside bomas. The bomas sampled formed a tapho-nomic sequence, having been abandoned from one to40 years prior to fieldwork. This allowed study of dungdeposition and its decomposition through time.

Materials and Methods

Figure 1. Map of Kenya showing the Rombo area (southern Kenya)where fieldwork was conducted.

Figure 2. Sketch map of the inhabited boma (in 1999). This boma isoccupied by four families, one of which is Ole Koringo’s. At least onegate belongs to a married man. Married women build their huts nextto their husband’s gate. Note the cattle enclosure in the central partof the boma and the smaller enclosures for caprines and calves.

Figure 3. Topographic map of the study area showing the locationof the inhabited (IB) and the four abandoned bomas (AB1 to AB40).Elevation contours are in feet.

Fieldwork

One inhabited boma and four abandoned bomas werelocated by a Maasai elder (Ole Koringo) with whom K.Ryan had previously worked. Ole Koringo’s familyhad lived at these sites and he remembered being achild at one of them, living as an adult in two others,and a fourth where friends of his lived. All bomas arelocated at about the same elevation and within an areaof 8 km2 (Figure 3). Fifty to sixty head of cattle and50–60 head of sheep and goats (caprines) are typicallypenned every night in bomas in the study area. Thebomas were sketch-mapped using a compass and tape.The inhabited boma allowed for ethnographic obser-vations to be made and the abandoned bomas pro-vided an opportunity to study the taphonomy of theenclosure sediments. Information on the times of their

442 R. Shahack-Gross et al.

abandonment was estimated by Ole Koringo based onwell-remembered events (such as initiations of agesets). This information was partly corroboratedusing air photographs obtained from the Surveys ofKenya.

The four bomas studied had been abandoned forapproximately: 5 months (2 years in the 2001 season,hence termed AB1), 15–20 years (hence termed AB20),30 years (hence termed AB30), and 40 years (hencetermed AB40). Bomas in the study area were typicallyoccupied for some 10–15 years. Abandonment of allsites was explained as due to ‘‘a need for change’’,except for AB1 that was abandoned following floodingby the Rombo River in December 1998. According toOle Koringo, none of the bomas was abandoned as aresult of the dung deposits being too deep, or due toparasite infestations (contra Western & Dunne, 1979).In addition, none of these bomas was burned afterabandonment (contra Western & Dunne, 1979).

Sampling locations within bomas included the cen-tral cattle enclosure and one or more caprine enclo-sures, identified by Ole Koringo. For comparison, atleast two samples of regional sediment were collectedseveral metres away from boma fences taking care notto sample gate areas where animals also deposit dung.Samples were collected through excavation of 1�1 mtest pits. These were excavated to a depth of about20 cm below the contact between enclosure sediments(i.e. dung) and the regional sediments (i.e. soil). Thiscontact was fairly easily distinguishable in AB1 andAB20 (based on colour and structural differences suchas lamination), but quite hard to discern visually inAB30 and AB40. Test pits for regional sediment sam-pling were dug to approximately 30 cm below surface.All test pits were photographed, and thicknesses ofstratigraphic units were measured. Loose sedimentsamples from different layers (distinguished primarilyby colour) were collected in plastic bags. In addition,block sediment samples were collected, some of themby using PVC pipes. Blocks were tightly wrapped withpaper and masking tape.

Laboratory techniquesEmbedded block samples were prepared followingconventional procedures (e.g., Courty et al., 1989).Thin sections were observed using a Nikon polarizinglight microscope (Labophot2-pol) and described fol-lowing Bullock et al. (1985) and Courty et al. (1989).

The pH of bulk sediment samples was measuredusing a pH-meter (Metrohm 654) after preparation ofsaturated solutions in water (Thomas, 1996).

Mineralogical identifications were performed onbulk samples using Fourier Transform Infrared(FTIR) spectroscopy (MIDAC Corp., Costa Mesa,CA, U.S.A.). Spectra were obtained by mixing about0·1 mg of powdered sample with about 80 mg of KBr.Spectra were collected at 4 cm�1 resolution (for moredetails see Weiner et al., 1993).

In some cases when mineralogical identificationby FTIR was difficult, X-ray powder diffraction(XRD) and Energy Dispersive X-ray spectrometry(EDS) techniques were also used. XRD spectra werecollected with a Ru200 generator and a D-Max/Bdiffractometer (both of Rigaku Corp., Japan) operat-ing with Cu anode at 40 kV and 120 mA. Scans weremade between 2–60� 2-Theta at a scan speed of 1� perminute. Data were interpreted using the Jade 5 DataAnalysis Software (Materials Data Inc., Livermore,CA, U.S.A.). Elemental analyses using the EDS tech-nique were performed on either polished, uncovered,petrographic thin sections or on powdered samplesmounted in Buehler Ultra-Mount and polished. Thepolished samples were carbon coated and analysedwith a Jeol 6400 scanning electron microscope with anEDS Link (Oxford Instruments) operating system.Images were recorded in the back-scattered electron(BSE) mode.

Quantitative phytolith analyses were performed us-ing Albert et al.’s. (1999) method, except that the acidtreatment was done without heating, and organic mat-ter in organic-rich sediments was removed by ashing(550�C for 4 h) prior to acid treatment. The sampleswere homogenized prior to analysis using a mortar andpestle. They were lightly ground in order not to breakthe phytoliths and sieved through a 0·5 mm sieve. Thefraction smaller than 0·5 mm was used for counting. Incontrast to Albert et al. (1999) who only countedopaline particles that clearly resemble phytoliths, inthis study every opaline particle was counted. Thisincludes particles as small as 2·5 �m that show apinkish to purple colour under plane polarized light(PPL) and are isotropic under crossed polarized light(XPL). This approach was taken because grasses con-tain very small geometric shaped short cell phytoliths(Mulholland & Rapp, 1992). Counting was performedon 10 fields; 2 dense fields in the central area of theslide and 8 peripheral fields. This represents a weightedmean of the phytoliths on the slide. The counting wasperformed using a Nikon petrographic microscope (asabove), at 400� magnification.

Six samples, three from caprine enclosures and threefrom cattle enclosures were analysed for phytolithmorphologies. More than 100 phytoliths with consist-ent morphologies were counted in each sample anddivided into morphological groups (following Albert &Weiner, 2001; Mulholland & Rapp, 1992).

Radiocarbon dating was performed on soil humatesthat were precipitated from four samples of regionalsediments along a 1·5 m deep soil profile. The sedi-ments were sampled from the following depths belowsurface: 5 cm, 40 cm, 90 cm and 150 cm. The soilsamples were treated with 1 N HCl to dissolveinorganic carbonate, followed by extraction of humicacids with 0·1 N NaOH. The humic acids were precipi-tated after addition of 1 N HCl, washed and dried. Thehumics were burned to release CO2 that was AMSdated. The samples and targets were prepared at the

Geo-Ethnoarchaeology of Pastoral Sites 443

Figure 4. Photographs of sediment profiles in open test pits. (a) Regional sediment in the vicinity of the inhabited boma. (b) Organic-richsediments from the cattle enclosure in AB1. (c) Sediments of a caprine enclosure in AB20. Note an upper oxidized, organic-poor, layer and anorganic-rich layer below it. (d) Sediments of the cattle enclosure in AB20. Note laminated layer (bracketed) above compacted, hard, regionalsediment. (e) Sediments in a caprine enclosure in AB30. There is no visual evidence of an organic-rich layer. (f) Sediments of the cattle enclosurein AB40. Note structural similarity of (e) and (f) to (a) and the absence of dark organic matter in (e), and (f). Arrows indicate the contactsbetween enclosure sediments and underlying regional sediments. Scale bar: 10 cm. Profile (f) is divided into an upper, post-abandonment, layer,two enclosure layers below it and a lower layer of compacted regional sediment. The numbers indicate phytolith concentrations in each layer(in millions per 1 g of total sediment). Note the distinct difference in phytolith concentrations between the underlying regional sediment and theenclosure sediments directly above it.

444 R. Shahack-Gross et al.

Radiocarbon Laboratory of the Weizmann Institute(Israel), and the analyses were performed at theUniversity of Arizona, Tucson.

ResultsWe describe field observations first, followed bydescriptions of the sediment micromorphology,mineral composition and finally phytolith contents andmorphologies. In each section we compare the analysesof the enclosure sediments to those of regionalsediments.

Field observations

Regional sediments. Soils in the Rombo area developedon Kilimanjaro Pleistocene basalts. They are brown orred clays that are relatively fertile, have good drainage,and contain calcite accumulations in the lower soilhorizons (Touber, 1983). No stratigraphy wasobserved at the depths dug in this study (Figure 4a),except for an upper finely stratified layer in the AB1location. No visible organic matter was noted.

Abandoned bomas and their enclosure sediments. Therecently abandoned boma, AB1, was similar in struc-ture to the inhabited boma; huts, fences and enclosureswere well preserved (Figure 5a). All other bomas were

so degraded that they could not been identified exceptby someone from the area. No features could beobserved, except for occasional hearthstones (seeFigure 5b–d).

In AB1, occupied for approximately 15 years, maxi-mal thickness of sediments deposited in the enclosureswere less than 1 m in the cattle area, and more than 1 min the caprine enclosures. Overall, it was noted thatdung accumulation in cattle enclosures is relatively flat,while in caprine enclosures it forms heaps (Figure 6).Enclosure sediments were organic-rich, wet (oozing),with thick massive black layers (Figure 4b).

In AB20, one caprine heap was still somewhatvisible, with a total thickness of 30 cm. The profile wasdivided visually into an upper white blocky layer(10 cm), a black-brown massive mottled layer (10 cm),and a massive black layer (10 cm), overlying the brownregional sediment (Figure 4c). The profile of the cattleenclosure sediments was c. 15 cm thick, with an upperlayer of crumbly sediment (up to 10 cm) overlyinga yellow to black-brown (2–7 cm) laminated layer(Figure 4d).

In AB30 and AB40, enclosure sediment profiles werequite thin (from 3 cm to 18 cm), and uniformlyorganic-poor. Layers with slightly different shades ofgray and brown were occasionally observed and it wasnot clear which of these layers had previously con-tained dung. In addition, the contact between theassumed enclosure sediments and the regional sedi-ment below was often visually unclear (Figure 4e–f).

Overall, enclosure sediment profiles observed in testpits tend to thin with time and lose their distinctiveblack (organic) colour after 20 to 30 years.

Average pH measurementsAverage pH measurements show that regional sedi-ments are slightly alkaline (7·6�0·24, n=13). Organic-poor enclosure sediments are even more alkaline(8·0�0·39, n=10) and organic-rich enclosure sedi-ments are highly alkaline (8·8�0·57, n=10). Regionalsediments underlying enclosure sediments are moresimilar to organic-poor enclosure sediments than toregional sediments (8·1�0·75, n=9).

Figure 5. Sketch maps of the four abandoned bomas. In AB1all features were mapped. In AB20, AB30 and AB40 only thecaprine enclosures that were sampled are shown. Note that not allconcentrations of hearth stones are drawn.

MicromorphologyTable 1 summarizes the micromorphological observa-tions of regional and enclosure sediments. It includesdescriptions of the microstructure, the fine fraction,and the distribution of the coarse fraction with respectto the fine fraction, termed related distribution. Inall samples, the coarse fractions are similar in com-position, sorting, and roundness. Therefore, the coarsefraction is described only once, in the regionalsediments section. The notes section includes descrip-tions of secondary features. Terminology followsBullock et al. (1985).

Geo-Ethnoarchaeology of Pastoral Sites 445

Regional sediments. The five embedded blocks pre-pared from sediments outside boma perimeters are allsimilar (i.e. controls; Figure 7a–b). They are composedof clay, grains of basalt and basalt-derived minerals(i.e. feldspars, pyroxenes and iron oxides, the latterbeing weathering products) and occasional pedogeniccarbonate nodules. The basalt and basalt-derivedgrains are randomly distributed in the clayey ground-mass (i.e. a porphyric related distribution). The finefraction is not oriented (i.e. an undifferentiatedbirefringence, or, b-fabric). Microstructures observedinclude poorly developed angular blocky, granular andcrumbly structures. The upper part of the sedimentsfrom AB1 shows laminated and crack structures withsome carbonate impregnation of the micromass andalong voids. These features result from flooding of thesite by the Rombo River.

Figure 6. Photographs of AB1 in May 1999 (a) and January 2001 (b). Note two caprine enclosures (elevated heaps, marked 1 and 2) and maincattle enclosure (flat, marked 3). A hut (4) observed in 1999 collapsed by 2001 and thornbush fences (5) observed in 1999 partially disintegratedin 2001. Fenced areas, where dung was not accumulating in large quantities, are marked by grass growth in 2001. Field of view: approximately20 m.

Enclosure sediments. AB1: Sediments from both cattleand caprine enclosures are dominated by organic

material that includes vegetal fibres and other recog-nizable plant tissues. The overall macroscopic structureis platy (Figure 7c). Planar voids are dominant withtraces of infilling by acellular decomposed organicmatter. The microstructure is laminated (Figure 7d).The fine fraction is mostly acellular organic matter thatsurrounds the coarse fraction or forms braces that linkthe coarser units (termed chitonic and gefuric relateddistributions, respectively; Table 1). The fine fractionmay contain spherulites. They are more abundant incaprine enclosure sediments than in cattle enclosuresediments. The spherulites and other carbonatic formsappear as birefringent areas in the groundmass (i.e. acrystallitic b-fabric; e.g., Figure 8c). The upper fewcentimetres in the regional sediment below theorganic-rich sediments is compacted due to livestocktrampling and has accumulations of carbonates in andalong voids (Figure 7e–f). These sometimes alsoimpregnate the clayey groundmass. This compactedstructure was observed in all sampled profiles that

446 R. Shahack-Gross et al.

T

able

1.M

icro

mor

phol

ogic

alde

scri

ptio

nsof

regi

onal

and

encl

osur

ese

dim

ents

.R

efer

tote

xtfo

rte

rmin

olog

y

Con

text

(sam

ple

nos)

Mic

rost

ruct

ure

Rel

ated

dist

ribu

tion

Fin

efr

acti

onN

otes

Reg

iona

lse

dim

ents

Rep

rese

ntat

ive

(KE

N-1

9,31

,13

3,18

1,22

5)M

ostl

ygr

anul

aran

dcr

umb

wit

hco

mpo

und

pack

ing

void

s.A

lso

poor

lyde

velo

ped

angu

lar

bloc

kyst

ruct

ure.

Oth

ervo

ids

are

vugh

s,ch

ambe

rsan

dch

anne

ls.

Ope

nto

clos

edpo

rphy

ric.

Lig

htto

dark

brow

ncl

ayw

ith

undi

ffer

enti

ated

b-fa

bric

.C

oars

efr

acti

on(c

.10–

20%

byar

ea,

ofse

dim

ent)

incl

udes

20�m

to4

mm

grai

nsof

basa

lt,

feld

spar

san

dol

ivin

e,po

orly

sort

edan

dan

gula

r.O

ccas

iona

lpe

doge

nic

carb

onat

eno

dule

s(u

pto

200

�m),

slug

skel

etal

gran

ules

and

poss

ible

eart

hwor

mor

gani

cca

sts

(up

to30

0�m

).C

rack

and

plat

ym

icro

stru

ctur

eat

top

ofsa

mpl

es31

and

181

due

toflo

odin

gev

ent,

toge

ther

wit

hcr

ysta

lliti

cb-

fabr

ic(c

arbo

nate

prob

ably

orig

inat

esfr

omw

ithi

nbo

ma

sedi

men

ts).

Cat

tle

encl

osur

epr

ofile

sA

B1

(KE

N-2

5)U

pper

(7cm

):C

ompa

cted

orga

nic

mat

ter.

Pla

tyst

ruct

ure,

plan

arvo

ids,

som

evu

ghs

and

chan

nels

.L

amin

ated

mic

rost

ruct

ure.

Por

phyr

ic.

Dar

kbr

own

orga

nic

mat

ter,

mos

tly

undi

ffer

enti

ated

b-fa

bric

,ar

eas

ofcr

ysta

lliti

cb-

fabr

ic.

Few

dung

sphe

rulit

es.

Veg

etal

mat

ter

(50

�mto

6m

mlo

ngpa

rtic

les)

form

sth

em

ain

part

(�30

%)

ofth

eco

arse

frac

tion

.L

arge

pass

age

feat

ures

.

Mid

dle

(7·5

cm):

Bio

turb

ated

,gr

anul

arst

ruct

ure,

com

poun

dpa

ckin

gvo

ids

and

vugh

s.

Por

phyr

ic.

Dar

kbr

own

tobl

ack

clay

wit

hpo

ssib

leor

gani

cst

aini

ng.

Und

iffer

enti

ated

and

crys

talli

tic

b-fa

bric

s.

Few

orga

nic

fibre

s,m

ostl

yin

the

uppe

r2–

3cm

.M

any

min

eral

grai

nsap

pear

tobe

ina

proc

ess

ofal

tera

tion

/dis

solu

tion

.A

reas

ofcl

ayey

grou

ndm

ass

impr

egna

ted

byca

rbon

ate.

Low

er:

Com

pact

edre

gion

alse

dim

ent,

sub-

angu

lar

bloc

kyan

dcr

umb

stru

ctur

es.

Pla

nar

and

com

poun

dpa

ckin

gvo

ids,

som

ech

anne

lsan

dcr

acks

.

Por

phyr

ic.

Bro

wn

clay

,cr

ysta

lliti

can

dun

diff

eren

tiat

edb-

fabr

ics.

AB

20(K

EN

-42,

43,

Upp

er(d

ivid

edin

tola

yers

aan

db)

:no

tor

ient

ed;

KE

N-1

08)

Lay

era

(1·7

cm):

Net

-lik

esp

ongy

stru

ctur

e,vu

ghs

and

cham

bers

.G

efur

ic.

Gre

yto

yello

wor

gani

cm

atte

ran

dca

rbon

ate,

crys

talli

tic

b-fa

bric

.

Hig

hly

frag

men

ted,

deco

mpo

sed,

vege

tal

mat

ter

(up

to1·

2m

mlo

ng).

Few

orga

nic

copr

olit

es.

Phy

tolit

hsco

mpr

ise

abou

t20

%of

fine

frac

tion

.

Lay

erb

(1·2

cm):

Pla

tyst

ruct

ure,

plan

arvo

ids.

Gef

uric

tocl

osed

porp

hyri

c.G

rey,

yello

wan

dbr

own,

clay

,ca

rbon

ate

and

orga

nic

mat

ter.

Cry

stal

litic

b-fa

bric

.

Smal

lar

eas

cont

ain

few

dung

sphe

rulit

es.

Phy

tolit

hsco

mpr

ise

abou

t30

–50%

ofth

efin

efr

acti

on.

Veg

etal

fibre

sco

mpr

ise

abou

t20

%of

coar

sefr

acti

on.

Low

er:

Com

pact

edre

gion

alse

dim

ent,

sub-

angu

lar

bloc

kyan

dbr

idge

d/pe

llicu

lar

ped

stru

ctur

es.

Com

poun

dpa

ckin

gan

dpl

anar

void

s,vu

ghs

and

chan

nels

.

Clo

sed

porp

hyri

c.L

ight

toda

rkbr

own

clay

and

carb

onat

e,un

diff

eren

tiat

edan

dcr

ysta

lliti

cb-

fabr

ics.

Som

evo

ids

coat

edw

ith

carb

onat

e,fe

ww

ith

clay

.

Geo-Ethnoarchaeology of Pastoral Sites 447

Tab

le1.

Con

tinu

ed

Con

text

(sam

ple

nos)

Mic

rost

ruct

ure

Rel

ated

dist

ribu

tion

Fin

efr

acti

onN

otes

AB

30(K

EN

-53,

224)

Upp

er:

Pos

tab

ando

nmen

tse

dim

ent,

gran

ular

stru

ctur

e.P

orph

yric

.L

ight

toda

rkbr

own

clay

,un

diff

eren

tiat

edb-

fabr

ic.

Roo

tac

tion

?

Mid

dle

(4to

13cm

):D

ecom

pose

den

clos

ure

sedi

men

t,an

gula

rto

sub-

angu

lar

bloc

kyst

ruct

ure,

gran

ular

and/

orcr

umbl

yat

plac

es.

Pla

nar

and

com

poun

dpa

ckin

gvo

ids

and

vugh

s.F

aint

mic

rola

min

ated

undu

lati

ngst

ruct

ures

.

Ope

npo

rphy

ric.

Shad

esof

brow

nan

dbl

ack,

clay

and

smal

lfr

agm

ents

ofor

gani

cm

atte

r.C

ryst

allit

icb-

fabr

icdu

eto

orga

nics

.

Lar

geam

ount

ofph

ytol

iths

.A

reas

ofla

min

arar

rang

emen

tof

blac

kor

gani

cpa

rtic

les.

Few

carb

onat

eno

dule

s.

Low

er:

Com

pact

edre

gion

alse

dim

ent

(�4

cm)

and

belo

wit

regi

onal

sedi

men

t.

Clo

sed

porp

hyri

c.L

ight

toda

rkbr

own

clay

and

carb

onat

e,un

diff

eren

tiat

edan

dcr

ysta

lliti

cb-

fabr

ics.

Lit

tle

carb

onat

ein

fillin

g/im

preg

nati

onof

void

s/gr

ound

mas

s.

AB

40(K

EN

-69)

Upp

er(5

cm):

Hig

hly

frag

men

ted

gran

ules

(�0·

5m

min

diam

eter

)of

encl

osur

ese

dim

ent

insu

b-an

gula

rbl

ocky

and

gran

ular

stru

ctur

es.

Pla

nar

and

com

poun

dpa

ckin

gvo

ids,

chan

nels

,cr

acks

and

vugh

s.D

isti

ncti

vem

icro

lam

inat

edun

dula

ting

stru

ctur

es.

Ope

npo

rphy

ric.

Gre

y,or

ange

,br

own

clay

,or

gani

cs,

opal

and

carb

onat

e.C

ryst

allit

icb-

fabr

icdu

eto

orga

nics

.

Roo

tac

tion

?

Low

er:

Mod

erat

ely

frag

men

ted

gran

ules

(�6

mm

indi

amet

er)

ofen

clos

ure

sedi

men

tin

sub-

angu

lar

bloc

kyst

ruct

ure.

Pla

nar

void

sdo

min

ate.

Dis

tinc

tive

mic

rola

min

ated

undu

lati

ngst

ruct

ures

.

Ope

nP

orph

yric

.G

rey,

oran

ge,

brow

ncl

ay,

orga

nics

,op

alan

dca

rbon

ate.

Cry

stal

litic

b-fa

bric

due

toor

gani

cs

Few

dung

sphe

rulit

es,

few

carb

onat

eno

dule

s,sl

ugsk

elet

algr

anul

esan

dpo

ssib

leea

rthw

orm

cast

s.

Cap

rine

encl

osur

epr

ofile

sA

B1

(KE

N-2

2)V

eget

alm

atte

rin

apl

atey

and

sub-

angu

lar

bloc

kyst

ruct

ures

,pl

anar

void

sdo

min

ate

but

also

vugh

s,ch

anne

lsan

dsi

mpl

epa

ckin

gvo

ids.

Chi

toni

c-ge

furi

c.O

rang

e-co

lour

edor

gani

can

dpo

ssib

lyor

gano

-met

allic

mat

ter

and

dung

sphe

rulit

es,

crys

talli

tic

b-fa

bric

.

Voi

dsco

ated

wit

ham

orph

ous

orga

nic

mat

ter.

Lar

gepa

ssag

efe

atur

es.

Few

Ca-

oxal

ate

drus

es.

Man

yph

ytol

iths

.

448 R. Shahack-Gross et al.

Tab

le1.

Con

tinu

ed

Con

text

(sam

ple

nos)

Mic

rost

ruct

ure

Rel

ated

dist

ribu

tion

Fin

efr

acti

onN

otes

AB

20(K

EN

-33,

39,

104)

Upp

er(1

3cm

):M

ostl

yin

orga

nic

wit

hco

mpl

exm

icro

stru

ctur

ein

clud

ing

angu

lar

bloc

ky,

vugh

yan

dcr

umb

stru

ctur

es.

Ped

sap

pear

tobe

com

pose

dof

wel

ded

dom

ains

ofca

rbon

ate,

opal

and

orga

nic

mat

ter,

wit

hdi

stin

ctiv

em

icro

lam

inat

edun

dula

ting

stru

ctur

es.

Oth

erst

ruct

ures

are

net-

like

and

opal

ine

lam

inae

/‘‘po

cket

s’’.

Lar

gepl

anar

void

s,ch

anne

ls,

cham

bers

and

vugh

s.

Upp

er,

alm

ost

com

plet

ely

min

eral

:op

enpo

rphy

ric.

Low

er,

cont

aini

ngso

me

vege

tal

mat

ter:

gefu

ric/

chit

onic

.

Gre

yto

yello

wis

hbr

own

and

few

blac

kar

eas,

com

pose

dof

carb

onat

e,or

gani

cm

atte

r,op

alan

dcl

ay.

Cry

stal

litic

b-fa

bric

.

Lar

ge(�

1–2

�m)

carb

onat

iccr

ysta

ls,

also

carb

onat

icpe

doge

nic

nodu

les.

Few

larg

eve

geta

lfib

res

and

seed

coat

frag

men

ts(u

pto

1·5

mm

long

)in

the

uppe

rpa

rt,

but

abou

t40

%of

fibre

s,se

edco

ats

and

othe

rpl

ant

tiss

ues

(up

to8

mm

long

)in

low

erpa

rt.

The

orga

nic

mat

ter

inth

elo

wer

part

ism

oder

atel

yw

ell

pres

erve

d(o

rang

eto

brow

nin

PP

L,

oran

geto

blac

kin

XP

L).

Few

Ca-

oxal

ate

drus

es.

Mid

dle

(10

cm):

Mos

tly

orga

nic

wit

hco

mpl

exm

icro

stru

ctur

ein

clud

ing

angu

lar

bloc

ky,

plat

yan

dsp

ongy

stru

ctur

es.

Lar

gepl

anar

void

s,vu

ghs

and

sim

ple

pack

ing

void

sbe

twee

nve

geta

lfr

agm

ents

.

Gef

uric

/chi

toni

c.D

ung

sphe

rulit

esco

mpr

ise

mos

tof

the

fine

frac

tion

,w

ith

som

eca

rbon

ate

and

amor

phou

sor

gani

cm

atte

r.C

ryst

allit

icb-

fabr

ic.

Veg

etal

mat

ter

ingo

odst

ate

ofpr

eser

vati

on(i

.e.,

high

bire

frin

gent

colo

urs

inX

PL

).

Low

er:

Mos

tly

regi

onal

sedi

men

t.U

pper

part

isa

mix

ture

ofre

gion

alse

dim

ent

wit

hve

geta

lm

atte

r,lo

wer

part

isco

mpa

cted

regi

onal

sedi

men

t.M

icro

stru

ctur

ein

clud

esco

mpa

cted

gran

ular

stru

ctur

ean

dpl

aty

atpl

aces

,bu

tal

socr

umb

and

spon

gyin

the

mix

edup

per

part

.C

ompo

und

pack

ing

void

san

dvu

ghs.

Clo

sed

porp

hyri

c.B

row

ncl

ay,

crys

talli

tic

and

undi

ffer

enti

ated

b-fa

bric

s.C

arbo

nate

impr

egna

tion

ofgr

ound

mas

san

dvo

idco

atin

gs.

Are

asof

carb

onat

eca

ppin

g/cr

usts

.

Geo-Ethnoarchaeology of Pastoral Sites 449

Tab

le1.

Con

tinu

ed

Con

text

(sam

ple

nos)

Mic

rost

ruct

ure

Rel

ated

dist

ribu

tion

Fin

efr

acti

onN

otes

AB

30(K

EN

-48,

223)

Upp

er(1

6cm

):G

ranu

lar

and

sub-

angu

lar

bloc

kyst

ruct

ures

,co

mpo

und

pack

ing

and

plan

arvo

ids

and

vugh

s.F

ewpe

dssh

owm

icro

lam

inat

edun

dula

ting

stru

ctur

es.

Por

phyr

ic.

Gre

y,ye

llow

,or

ange

and

brow

ncl

ayan

dsi

lt,

incl

udin

gca

rbon

ate

and

opal

.C

ryst

allit

icb-

fabr

icbu

tal

soar

eas

ofst

riat

edb-

fabr

icw

here

orga

nic

fibre

spr

eser

ve.

Dun

gsp

heru

lites

are

rare

and

mic

rola

min

ated

undu

lati

ngst

ruct

ures

are

fain

t.L

arge

pass

age

feat

ures

and

som

eor

gani

cca

sts.

Few

area

sim

preg

nate

dw

ith

carb

onat

e.T

hefr

eque

ncy

ofpe

dsla

rger

than

1m

min

diam

eter

rise

sbe

low

5cm

dept

h.O

vera

ll,en

clos

ure

sedi

men

tis

poor

lypr

eser

ved.

Low

er:

com

pact

edre

gion

alse

dim

ent.

Pla

tyan

dsu

b-an

gula

rbl

ocky

stru

ctur

es.

Clo

sed

porp

hyri

c.B

row

ncl

ay,

undi

ffer

enti

ated

and

crys

talli

tic

b-fa

bric

s.C

arbo

nate

impr

egna

tion

mos

tly

alon

gvo

ids.

AB

40(K

EN

-63)

Ang

ular

bloc

kyst

ruct

ure,

gran

ular

orpl

aty

atpl

aces

.P

lana

ran

dco

mpo

und

pack

ing

void

san

dvu

ghs.

Dis

tinc

tive

mic

rola

min

ated

undu

lati

ngst

ruct

ures

.

Ope

npo

rphy

ric.

Bro

wn,

grey

,or

ange

and

rest

rict

edar

eas

ofgr

eeni

shye

llow

,cl

ay,

carb

onat

e,op

alan

dor

gani

cm

atte

r.C

ryst

allit

icb-

fabr

icdu

eto

larg

eam

ount

s(>

50%

)of

dung

sphe

rulit

es.

Car

bona

tic

and

orga

nic

stai

ning

,al

soa

few

pedo

geni

cca

rbon

ate

and

phos

phat

eno

dule

s.F

ewC

a-ox

alat

edr

uses

.P

ossi

ble

eart

hwor

mca

sts

and

apo

ssib

ledu

ngbe

etle

pelle

t(r

ound

ed,

c.2

cmin

diam

eter

,m

ade

ofve

ryw

ell

pres

erve

dve

geta

lm

atte

r).

Ped

sar

em

ore

frag

men

ted

atth

eup

per

6cm

ofth

epr

ofile

and

pack

edin

asu

b-an

gula

rbl

ocky

stru

ctur

e,w

hile

bello

w6

cmde

pth,

peds

are

larg

eran

dar

rang

edin

abl

ocky

-pla

tyst

ruct

ure.

450 R. Shahack-Gross et al.

Figure 7. Photographs of thin sections (a, c, e) and representative photomicrographs (b, d, f) of areas in these thin sections. The scale bar in(c) applies to (a) and (e). The scale bare in (d) applies to (b) and (f). (a) Regional sediment outside AB20 (sample KEN-133). Note granularstructure and angular blocks. (b) Photomicrograph of same sample in PPL. Note pyroxene grain (1), feldspar grains (2) and overallgranular/crumb structure. Opaque grains are iron oxides (3). (c) Organic-rich enclosure sediment from AB1 (sample KEN-25a). Note platystructure. (d) Photomicrograph of same sample in PPL. Note vegetal fibres (4) forming a microlaminated structure. (e) Compacted regionalsediment underlying caprine enclosure sediments in AB20 (sample KEN-104b). Note platy appearance of compacted area (bracketed, c.f.,Fig. 7a) (f) Photomicrograph of the compacted area from the same sample, in PPL. Note compressed granules and planar voids in the overallplaty structure (c.f., Fig. 7b). Groundmass and voids are impregnated and infilled with carbonate.

included regional sediments overlain by enclosure sedi-ments (Table 1).

AB20: The upper parts of the caprine enclosuresediments from AB20 are organic-poor and are mostlycarbonatic with a complex microstructure (describedbelow). Spherulites are very abundant throughout theenclosure sediment profile. The middle part of theprofile is still rich in organic matter. The lowermostpart of the caprine enclosure sediment is a mixture oforganic matter and regional sediment.

The cattle enclosure sediments in AB20 are disturbedby root action in their upper 15 cm. Below the rootaffected sediment, there is a laminated layer of organic-rich enclosure sediments. The organic matter is ina poor state of preservation, indicated by its lowbirefringence.

In the organic-poor enclosure sediments in AB20 wefirst observed a microlaminated undulating structurethat characterizes degraded enclosure sediments

(Figure 8a). This structure is composed of alternatinglaminae of acellular organic matter and opal phyto-liths. The laminae are about 10–30 �m thick, and areon the average continuous for distances up to 2 mm,and generally meandering (i.e. not straight). This struc-ture is not mentioned in Bullock et al. (1985), and maybe similar to a structure observed by Wattez et al.(1990) but not elaborated on. Based on the micro-morphological observations of the taphonomicsequence described in this study, we agree with Wattezet al. (1990) that the microlaminated structure resultsfrom trampling by livestock. Such trampling promotesre-organization of plant fibres found in herbivorousdung to form a platy macrostructure, sub-parallel tothe sediment surface. Furthermore, we suggest that theundulations result from volume changes due to thedegradation of organic matter, although phytolithsand small quantities of acellular organic matter are stillpreserved.

Geo-Ethnoarchaeology of Pastoral Sites 451

AB30: Enclosure sediments could be identified bythe presence of isolated pockets of the undulatingmicrolaminated structure. Hence enclosure sedimentscould be pinpointed. The sediments did not includemany carbonate minerals, and spherulites were rare inthe caprine enclosure sediments.

AB40: Cattle and caprine enclosure sedimentsfrom AB40 are similar in their structure and texturesbut differ in their mineralogical composition(Figure 8b–e). They have an overall angular to sub-angular blocky structure with areas of granular orplaty structures (Figure 8f). Both show the charac-teristic microlaminated undulating structure. The dif-ference between them is that the caprine enclosuresediments contain large amounts of carbonates,especially in the form of spherulites, when comparedto the cattle enclosure sediments (cf. Figure 8b–c withFigure 8d–e). Due to the high frequency of opalphytoliths in both types of enclosure sediments, they

appear to be isotropic in XPL (Figure 8e), except forthe high birefringence of carbonate minerals andsome preserved organic fibres (Figure 8c, e). Second-ary features include a few yellow and green (asobserved in PPL) pedogenic phosphatic nodules thatare isotropic in XPL.

Figure 8. Organic-poor caprine and cattle enclosure sediments. (a) Photomicrographs of an ‘‘undulating’’ microlaminated structure inorganic-poor caprine enclosure sediments from AB20 (sample KEN-39). Note alternating micro-layers of opal (transparent) and degradedorganic matter (opaque). Overall groundmass in carbonate. (b) Photomicrograph from sample of caprine enclosure sediment from AB40(KEN-63) in PPL and (c) in XPL. Note dung spherulites appear as birefringent granules in (c). (d) Photomicrograph of cattle enclosuresediment AB40 (sample KEN-69b) in PPL and (e) in XPL. Note that in comparison with the caprine enclosure sediment (b and c), there arefewer birefingent area in the cattle enclosure sediment under XPL. In both cases, the opaque groundmass in XPL is mostly isotropic composedof large numbers of opal phytoliths. (f) Photograph of thin section of caprine enclosure sediment in AB40 (sample KEN-63c). Note sub-angularblocky, granular and platy structures.

Mineralogical characterization of bulk samplesTable 2 shows the minerals present in regionaland enclosure sediments. Regional sediments in thestudy area are largely composed of clay minerals(Figure 9a). The XRD patterns of several samplesshow that clays are mostly of the kaolinite group.Vermiculite is probably also present. EDS analysesshow that Si and Al are the major components in theclay mixture (Si constitutes 19–27% and Al consti-tutes 11–18%). Minor components include Mg andNa (1–2% each), and also P, K and Ca (usually less

452 R. Shahack-Gross et al.

than 1% each). Also present are Fe-Ti oxides that areincluded in the clayey groundmass (Fe constitutes8–12% and Ti constitutes 1–2%). Based on EDSanalyses, feldspars observed in the micromorphologi-cal thin sections belong to the plagioclase group. Thepresence of the pyroxene diopside (CaMgSi2O6) wasalso inferred using the EDS.

Organic-rich enclosure sediments are composed ofa mixture of clay, monohydrocalcite (CaCO3 . H2O),organic material and its decomposition products,notably ammonium sulfate (Figure 9b). Mono-hydrocalcite is found in higher concentrations incaprine compared with cattle enclosure sediments(Table 2).

Organic-poor enclosure sediments are composedmainly of clay and opal (Figure 9c) and most caprineenclosure sediments include monohydrocalcite as well.Some samples contain Mg-rich calcite (Figure 9d).EDS analyses of the phosphatic nodules observedin the oldest enclosure sediments yielded a non-stoichiometric ratio of Ca and P of 1·16, suggestingthat the mineral is either amorphous Ca-phosphate ora non-stoichiometric dahllite (Ca5(PO4,CO3)3(OH))(LeGeros, 1991).

Table 2. Minerals present in regional and enclosure sediments. The average weight % of the acid insoluble fraction (AIF) is shown for each category

Category Sub-category Major components Minor components Average weight % of AIF

Regional sediments Clay (kaolinite). Vermiculite, plagioclase,olivine, iron oxides, sodiumnitrate, possiblecarbonates, opal,unidentified peaks at 1010,754 and 746–8 cm�1.

89·3�5·87 (n=15)

Cattle enclosure sediments Organic rich Clay, organic matter. Monohydrocalcite. 67·1�21·55 (n=3)After ashing: 46·1�0·07 (n=2)

82·8�9·81 (n=16)Organic poor Clay, opal. Monohydrocalcite, organic

matter.

Caprine enclosure sediments Organic rich Clay, monohydrocalcite,ammonium sulfate, organicmatter.

Mg-rich calcite,unidentified peaks at 1155,754 and 615 cm�1.

55·6�16·92 (n=5)After ashing: 32·6�11·95 (n=4)

71·3�15·34 (n=11)Organic poor Clay, monohydrocalcite,

opal, Mg-rich calcite.Organic matter, possibleCa-oxalate, unidentifiedpeaks at 879, 688 and574 cm�1.

Overlain regional sediments Clay. Carbonate, opal, sodiumnitrate, unidentified peaksat 688 and 754 cm�1.

87·0�6·36 (n=14)

Quantitative phytolith analysesPhytolith preservation is an important issue in alka-line soils, as opal solubility increases with elevatedpH (Karkanas et al., 2000). Many of the phytolithsstudied seem to be weathered, as indicated by theirfragmentary nature and pitted surfaces. In order to

determine whether or not phytoliths may preserve forlong periods of time even in these alkaline soils, afresh soil profile of 1·5 m was exposed on the bank ofa small creek draining into the Rombo River. Figure10 shows the number of phytoliths per gram of totalsediment along this profile and the radiocarbon datesobtained from some of these samples. As there isno significant reduction in phytolith concentrationswith increasing depth (i.e. age) we conclude thatphytoliths can preserve for thousands of years inthese alkaline sediments. The 1180�40 (uncali-brated) age obtained for the sample from 5 cm belowsurface probably indicates the removal of themore recent part of the profile by river action, thusexposing a paleo-surface.

The overall error in phytolith concentrations wascalculated as the % standard deviation between dupli-cated or triplicated samples. For homogenized samplesit is �29% (n=5). This incorporates �9% (n=3) errorin counting, calculated by counting the same slidetwice. The remaining 20% is due mainly to weighingerrors and loss of material during the heavy liquidseparation procedure.

A consistent pattern was observed in the relativeweights of the six fractions produced during the heavyliquid separation procedure (Figure 11). In enclosuresediments most of the material is concentrated in the5th and 6th fractions (i.e. the light fractions contain-ing mostly opal and organic matter), whereas inregional sediments most of the material is concentratedin the 1st through 3rd fractions (i.e. denser fractionscontaining mostly feldspars and clay).

Geo-Ethnoarchaeology of Pastoral Sites 453

Phytolith concentrations from enclosure andregional sediments are shown in Figure 12. Concen-trations in regional sediments are all lower than 10million phytoliths per 1 g sediment and their range isquite restricted. The majority of samples have less than4 million phytoliths per gram sediment. On the otherhand, concentrations in enclosure sediments averagemore than 20 million phytoliths per gram sediment.The range of concentrations is very large and doesoverlap with that of the regional sediments.

Morphological analyses of phytoliths from cattleand caprine enclosure sediments show that they areindistinguishable (Figure 13). Almost all phytolithforms observed belong to grasses. Phytolith mor-phologies are therefore not useful for differentiatingbetween cattle and caprine enclosures.

DiscussionThis study shows that sediments from open-air animalenclosures in Maasai pastoral sites can be differenti-

ated from surrounding sediments, even 40 years afterabandonment, by using a suite of analytical tech-niques that characterize the structure, mineralogy andphytoliths of sediments.

Structural indicators of enclosure sedimentsModern enclosure sediments are composed of largequantities of organic matter and small amounts ofbiogenically produced materials (i.e. spherulites, phyto-liths). They can be visually distinguished from regionalsediments by their black colour and laminated struc-ture. Based on field observations, however, enclosuresediments cannot be visually differentiated fromregional sediments 20 to 30 years after site abandon-ment, because the organic matter disintegrates as aresult of oxidation. Detailed examinations of thinsections confirm that only very small amounts oforganic matter are preserved 30 years after site aban-donment. The bulk of the organic-poor enclosuresediments are composed of concentrations of the bio-genic minerals included in dung, together with clay.These contribute to an overall texture that is finergrained in enclosure than in regional sediments, anobservation also made by Brochier et al. (1992).

The platy structure characteristic of trampledorganic-rich enclosure sediments may be preserved inorganic-poor enclosure sediments. However, in mostcases it is lost at the macroscopic level, due to rootaction and bioturbation. This results in a blocky togranular structure that is similar to that of regionalsediments. Microscopically, however, the lamination istransformed into the unique microlaminated undulat-ing structure, which can be observed even in the oldestenclosures examined. It is this feature that allows us todifferentiate the base of the enclosure sediments fromunderlying regional sediments. In addition, we notethat the underlying regional sediment appears to becompacted due to trampling.

Overall, the structural indicators for organic-poorenclosure sediments are microlaminated undulatingstructures, a texture that is finer grained than inregional sediments, and the presence of com-pacted structures in the regional sediments underlyingenclosure sediments.

Figure 9. FTIR spectra of (a) Representative regional sedimentcomposed almost entirely of clay (the peaks at 1097 and 690 cm�1

are from an unidentified material). (b) Organic-rich sediment from acaprine enclosure sampled in AB1. The peaks at 1116 and 618 cm�1

are from ammonium sulfate. Monohydrocalcite, characterized by thesplit between 1490 and 1414 cm�1, produces the peaks at 875, 763,699 and 568 cm�1. The broad absorption around 1600 cm�1 is dueto organic matter. The rest of the peaks are from clay (cf. Fig. 9a)and an unidentified material in 1009 cm�1. (c) Laminated sedimentfrom the cattle enclosure in AB20. Opal is the main component ofthis sample (peaks at 1103, 796 and 473 cm�1). The sample alsocontains clay and organic matter. (d) Organic-rich sediment from acaprine enclosure sampled in AB20. The lack of splitting in the1400 cm�1 region, together with the peak at 876 cm�1 and the weakand broad absorption centred around 713 cm�1 indicates that themineral is Mg-rich calcite. It is probably in the process of recrystal-lization from monohydrocalcite. The sample also contains clay,organic matter and ammonium sulfate. The peak at 576 cm�1 isfrom an unknown material.

Mineralogical indicators of enclosure sedimentsThe biogenic minerals (carbonates and opal) present insmall amounts in fresh dung are concentrated withtime as organic matter is degraded. Carbonates appearin the form of spherulites and of calcite impregnatingthe groundmass and infilling voids. Opal is found inthe form of phytoliths, mostly from ingested grasses.The mineral monohydrocalcite, a relatively unstableform of calcium carbonate, was identified, mainly incaprine enclosures. Its preservation is probably due tothe alkaline nature of the sediments in the study area.Because of its high solubility it would not be expected

454 R. Shahack-Gross et al.

Figure 10. Phytolith concentrations as a function of depth in the deep soil section sampled in order to assess phytolith preservation in soils ofthe study area. The brackets indicate the counting error of �29% for each sample. Note the general similarity in phytolith concentrations alongthe soil profile, indicating that phytoliths do preserve in soils of the study area.

Figure 11. Average weight % of the fractions produced in the heavyliquid separation procedure, of regional sediments (filled squares,n=10) and organic-poor enclosure sediments (open circles, n=11).Note that light minerals (i.e. opal in fractions 5 and 6) are found inhigher amounts in enclosure sediments, indicating higher amounts ofphytoliths in enclosure sediments relative to regional sediments.

to preserve for very long. It may, however, dissolve andre-precipitate in the form of more stable minerals (seebelow). There is a rough correlation between theconcentration of monohydrocalcite in sediments andthe presence of microscopic dung spherulites. Thismay well indicate that dung spherulites, the exactmineralogical composition of which has not beendetermined to date (cf. Canti 1997, 1998, 1999), arecomposed of monohydrocalcite.

We note that organic-rich enclosure sediments fromAB20 and also organic-poor enclosure sediments fromAB40 contain Mg-rich calcite (Figure 9d). This mightbe a result of dissolution of monohydrocalcite andre-precipitation of calcite in an environment rich inMg. Possible sources for Mg are, first, weathering ofprimary minerals such as diopside, second, weatheringof vermiculite, and third, Mg that is found in animalurine (Alfrey, 1985). We also note partial transforma-tion of carbonates to Ca-phosphate. The source ofthe phosphate is probably degrading organic matter.Formation of Ca-phosphates is the first step along awell-defined diagenetic pathway in which, dependingupon the paleochemical environment in the sediment,minerals either dissolve and leave the system, or trans-form into a more stable form (Karkanas et al., 2000).We can thus expect that ancient caprine enclosuresediments in particular, will contain relatively stablephosphate minerals derived from the primary solublemonohydrocalcite. In contrast, we expect that cattleenclosures will have either none or low quantities of

Geo-Ethnoarchaeology of Pastoral Sites 455

Figure 12. Phytolith concentrations in total sediments from regional and enclosure sediments. Note the restricted and low phytolithconcentrations in regional sediments relative to enclosure sediments which have a large range of phytolith concentrations. Refer to text forinterpretation.

Figure 13. Percentage of phytolith morphological groups in cattle (white) and caprine (black) enclosure sediments. ppp refers to parallelpipedmorphology, sc refers to short cells. Note the overall similarity in phytolith morphologies between cattle and caprine enclosure sediments. Fordetailed description of phytolith types, refer to Albert & Weiner (2001).

456 R. Shahack-Gross et al.

phosphate minerals because they are relatively poor inprimary monohydrocalcite.

We have further noted that neither monohydrocal-cite nor spherulites were found in the caprine enclo-sures sampled at AB30, and the caprine enclosuresampled in AB20 in the 2001 season. In contrast,enclosure sediments from AB40 are relatively wellpreserved and do contain monohydrocalcite. AB20 andAB30 are located close to the river while AB40 islocated in an elevated area. Enclosure sediments maytherefore be better preserved in sites that are locatedaway from water sources.

Dissolved monohydrocalcite is the main source ofthe carbonate that percolates down the enclosure sedi-ment profile and impregnates the underlying regionalsediment. The presence of high amounts of carbonateis the reason why organic-poor enclosure sedimentshave elevated pH values, and also why the underlyingregional sediments have a pH value slightly above thatof other regional sediments.

Overall, the mineralogical indicators for enclosuresediments are opal (phytoliths), monohydrocalcite,and its more stable derivatives; Mg-rich calcite andCa-phosphate minerals.

Phytolith concentration as an indicator of animal enclo-sures

Phytoliths may preserve for thousands of years inregional sediments of the study area. The phytolithsare pitted, however, in all of the samples analysed inthis study, indicating that they have partially dissolved.We also note that organic-rich enclosure sedimentscontain larger quantities of small opaline fragmentsthan organic-poor enclosure sediments. Furthermore,organic-poor enclosure sediments contain opaline par-ticles that appear to be halves of original dumb-bell-shaped phytoliths. These observations suggest thatphytoliths in sediments of the study area undergodissolution resulting in assemblages that are biasedtowards large and thick phytoliths. Because opal isrelatively soluble above pH 8·5, and because such pHlevels occur primarily in organic-rich sediments (prob-ably due to high levels of ammonia from urine), wesuggest that this dissolution occurs in the alkalineenvironment of the organic-rich sediments. Note thathigh pH values (8·8�0·57) were measured for thewater in equilibrium with the organic-rich sediments.Phytolith dissolution greatly diminishes after the pHdeclines to the regional levels (7·6�0·24), in a matterof 30 years or so after abandonment. This scenario isconsistent with the remaining phytoliths surviving forthousands of years in organic-poor sediments.

Phytolith concentrations in most enclosure sedi-ments are higher than in regional sediments (Figures 12and 4f). Concentrations in regional sediments (includ-ing underlying ones) are quite low with the majority ofsamples (21 out of 25) having less than 2 million

phytoliths in 1 g sediment. Regional sediments withmore than 2 million phytoliths in 1 g sediment areprimarily from samples taken close to boma gates.These samples may have contained small amounts ofdung and hence higher amounts of phytoliths, becauseof livestock passage through the gate.

The range of phytolith concentrations in enclosuresediments is quite large with the majority of samples(25 out of 30) having concentrations above 2 millionphytoliths in 1 g sediment. Samples with low concen-trations of phytoliths may result from mistakenidentification of the contact between enclosure andunderlying regional sediments. We also note that con-centrations of phytoliths in enclosure sediments mayonce have been higher as several taphonomic processeswill result in their dilution. These processes may be theaddition of clay into the enclosure sediment as a resultof clay movement down the sediment profile (i.e.clay elluviation) and/or in situ clay formation fromweathered primary minerals.

Phytolith counts from other boma features, housesand hearths, show that sediments from these featureshave concentrations of phytoliths similar to regional,rather than enclosure sediments. Note that in sedi-ments containing abundant soluble minerals, such asmonohydrocalcite in caprine enclosure sediments,phytoliths will become more concentrated due todiagenetic loss of the carbonates.

Phytolith morphologies (Figure 13) show that cattle,and sheep and goat, enclosures cannot be distinguishedbased on the phytoliths present in their dung. Bothassemblages are dominated by grass phytoliths. Cattleand sheep are grazers while goats are mostly browsers,but feed on grasses as well. The result of penning sheepand goats together is that caprine dung deposits mostlycontain grass phytoliths, similar to those produced bycattle.

In conclusion, concentrations above 2 millionphytoliths in 1 g of sediment are indicative of enclosuresediments in the study area. However, because regionaland enclosure sediments overlap in the low concen-tration range, it might be better to only regard signifi-cantly higher concentrations of phytoliths, than thosein regional sediments, as a definitive indicator ofancient enclosure sediments.

Other indicators

Previous studies noted several other indicators of live-stock enclosures. These include shed milk teeth oflambs and kids, goat hairs, insect remains, and diatomsthat originate from ingested water (Brochier et al.,1992; Goldberg & Whitbread, 1993). In addition,Wattez et al. (1990), Courty et al. (1991) and di Lernia(1998) report on preserved intact oval caprinepellets. We did not observe these materials and/orstructures in open-air sites with organic-poor enclosuresediments.

Geo-Ethnoarchaeology of Pastoral Sites 457

Implications for the study of site formationprocessesThis study provides information on early diageneticprocesses of organic-rich sediments. Karkanas et al.(2000) proposed that diagenesis in prehistoric cavesoccurs at the surface (i.e. during sediment deposition)or soon after burial. They argue that the driving forcesare water passage through the sediment and the pres-ence of large amounts of organic matter (bat guano, inthe case of prehistoric caves). Their model also suggeststhat diagenesis slows down considerably, soon afterorganic matter is degraded. But, they had very littleinformation about how long organic matter might taketo degrade. In this study we show that in open-air sites,almost all the organic matter is degraded within30 years or so after abandonment. Furthermore, weshow that amorphous Ca-phosphate is already formingin a site that had been abandoned for 40 years. If theKarkanas et al. (2000) model is correct, then perhapsthe changes that take place during the first 40 or soyears of site formation encompass most of the majorchanges that will take place during diagenesis. In thiscase, we can anticipate that the micromorphologicalfeatures, mineral distributions and phytolith concen-trations observed in this study, could all be potentiallypreserved in older, archaeological, sites. If this is true,then studying site formation processes using anethnoarchaeological approach that focuses on sedi-ment sampling in abandoned sites, may be extremelyvaluable for better understanding the archaeologicalrecord.

Practical suggestionsIn practice, we suggest that in choosing an archaeologi-cal site for excavation, that may have been occupied bypastoralists, the archaeologist should select for settingsaway from local water sources, because carbonate andphosphate minerals will be better preserved in suchlocales. Coring the chosen site along a grid may beuseful for initial bulk mineralogical analyses in order tolocate animal enclosures. It is important to extend thegrid beyond the site’s perimeter to provide samples ofregional sediments that will serve as controls. Oncepotential enclosures are identified, it is suggested thattest pits be excavated and samples for phytolith analy-ses taken from the surface down the profile every 5 cm.They should cover all strata suspected to be enclosuresediments, as well as the sediments above and belowthem. Sediment samples may first be analysed using theheavy liquid procedure. The distribution of fractionweight % (as presented in Figure 11) is a useful meansof distinguishing enclosure sediments from regionalsediments, before actually counting phytoliths.Samples for micromorphology should be taken alongopen profiles, one above the other with a slight overlapat the boundaries between two samples (cf. fig. 3.9 inCourty et al., 1989). They should include all strata that

are suspected of being enclosure sediments, as well asthe sediments above and below them.

ConclusionsUsing a combination of micromorphological features,mineral distributions and phytolith concentrations, it ispossible to identify livestock enclosures in open-airsites. It is important to note that using one techniquealone is not sufficient for a definitive identificationof an enclosure and hence pastoralist occupation.In addition, it is important to compare suspectedenclosure sediments with regional, control sediments.

This study shows that enclosure sediments differmicromorphologically from regional sediments,because of the unique microlaminated undulatingstructure that they contain. Mineralogically, enclosuresediments differ from regional sediments by thepresence of minerals that are derived from livestockdung, such as monohydrocalcite. Where preservation ispoorer, we expect to find the more stable derivatives ofmonohydrocalcite, namely Mg-rich calcite and/orphosphatic minerals. These are especially apparent incaprine enclosure sediments. Enclosure sediments canalso be distinguished from regional sediments byphytolith contents. Concentrations are usually higherthan 2 million phytoliths in 1 g sediment, occurring inenclosure sediments. Phytolith morphologies cannot beused to differentiate cattle from caprine enclosures.

This study provides analytical tools with which toidentify ancient livestock enclosures in East Africansites. These will allow archaeologists to identifypastoral sites more securely and to address questionsrelating to pastoral expansions in the region. Inparticular, these methods will allow identification ofherding by local hunter-gatherer groups, and discrimi-nation of this practice from accumulation of domesticanimals for immediate consumption through trade orraiding. In addition, the same tools will be useful foridentifying penning of proto-domestic animals andthus the process of domestication of cattle, sheep andgoats in other contexts such as North Africa or theNear East.

This study demonstrates the utility of combiningethnographic fieldwork with quantitative laboratorystudies. In addition, a geo-ethnoarchaeologicalapproach enables us to ‘‘calibrate’’ site formationprocesses, and thus allows control over the timing ofdiagenesis and a more precise interpretation of siteformation processes.

AcknowledgementsR.S.-G. is indebted to K. Ryan and Dr Karega-Munene, without whom fieldwork in Kenya couldnot have been carried out. We thank Ole Koringo,Lekatoo Ole Parmitoro, S. M. Kahinju, P. M. Kunoni,C. Ngange, K. Seberini and J. Ndungu. Their support

458 R. Shahack-Gross et al.

in all aspects of the work, good spirits and personaltouch are much appreciated. We thank W. R. Fitts forsurveying data, P. Goldberg and P. Karkanas for theircomments on the micromorphology, E. Boaretto andE. Mintz for help with the radiocarbon dating ofsamples, and R. M. Albert for help with phytolithidentifications and counting. S. Raz and F. Bernahelped with instrumentation and general advice. Weare also grateful to three anonymous reviewers fortheir comments.

Permission for the field study was given to R.S.-G.from the government of Kenya, in collaboration withthe National Museums of Kenya. Airphotos and topo-graphic maps were obtained from the Surveys ofKenya with the great help of Z. Otieno, the NationalMuseums of Kenya. This study was supported by anNSF grant (Doctoral Dissertation ImprovementGrant, award no. 57508), a Wenner-Gren Foundationgrant (Gr. 6663) and the Kimmel Center for Archaeo-logical Science, Weizmann Institute, to R.S.-G., and anIsrael Science Foundation grant to S.W.

ReferencesAkeret, O. & Rentzel, P. (2001). Micromorphology and plant

macrofossil analysis of cattle dung from the Neolithic lake shoresettlement of Arbon Bleiche 3. Geoarchaeology 16, 687–700.

Albert, R. M., Lavi, O., Estroff, L., Weiner, S., Tsatskin, A., Ronen,A. & Lev Yadun, S. (1999). Mode of occupation of Tabun Cave,Mt Carmel, Israel during the later Mousterian: a study of thesediments and phytoliths. Journal of Archaeological Science 26,1249–1260.

Albert, R. M. & Weiner, S. (2001). Study of phytoliths in prehistoricash layers using a quantitative approach. In (J. D. Meunier & F.Colin, Eds) Phytoliths, Applications in Earth Science and HumanHistory. A. A. Balkema Publishers, pp. 251–266.

Alfrey, A. C. (1985). Disorders of magnesium metabolism. In(D. W. Seldin & G. Giebisch, Eds) The Kidney: Physiology andPathophysiology. New York: Raven Press, pp. 1281–1295.

Ambrose, S. H. (1984). The introduction of pastoral adaptations tothe highlands of East Africa. In (J. D. Clark & S. A. Brandt, Eds)From Hunters to Farmers: The Causes and Consequences of FoodProduction in Africa. Berkeley: University of California Press,pp. 212–239.

Ambrose, S. H. (1998). Chronology of the Later Stone Age and foodproduction in East Africa. Journal of Archaeological Science 25,377–392.

Ambrose, S. H. (2001). East African Neolithic. In (P. N. Peregrine& M. Ember, Eds) Encyclopedia of Prehistory, volume 1: Africa.New York: Kluwer Academic/Plenum Publishers, pp. 97–109.

Andersen, K. B. (1978). African Traditional Architecture. Nairobi:Oxford University Press.

Banning, E. B. & Kohler-Rollefson, I. (1992). Ethnographic lessonsfor the pastoral past: camp locations and material remains nearBeidha, Southern Jordan. In (O. Bar-Yosef & A. Khazanov, Eds)Pastoralism in the Levant: Archaeological Materials in Anthropo-logical Perspectives. Chicago: Prehistory Press. Monographs inWorld Archaeology No. 10, pp. 181–204.

Blench, R. M. & Macdonald, K. C. (Eds) (2000). The Originsand Development of African Livestock: Archaeology, genetics,linguistics and ethnography. London: UCL Press.

Boschian, G. & Montagnari-Kokelj, E. (2000). Prehistoric shepherdsand caves in the Trieste karst (Northeastern Italy). Geoarchaeology15, 331–371.

Bower, J. (1991). The Pastoral Neolithic of East Africa. Journal ofWorld Prehistory 5, 49–82.

Brochier, J. E. (1983). Combustion et parcage des herbivoresdomestiques. Le point de vue du sedimentologue. Bulletin de laSociete Prehistorique Francaise 80, 143–145.

Brochier, J. E., Villa, P. & Giacomara, M. (1992). Shepherds andsediments: geo ethnoarchaeology of pastoral sites. Journal ofAnthropological Archaeology 11, 47–102.

Bull, I. D., Simpson, I. A., van Bergen, P. F. & Evershed, R. P.(1999). Muck ’n’ molecules: organic geochemical methods fordetecting ancient manuring. Antiquity 73, 86–96.

Bullock, P., Fedoroff, N., Jongerius, A., Stoops, G. & Tursina,T. (1985). Handbook for Soil Thin Section Description.Wolverhampton, England: Waine Research Publications.

Canti, M. G. (1997). An investigation of microscopic calcareousspherulites from herbivore dungs. Journal of ArchaeologicalScience 24, 219–231.

Canti, M. G. (1998). The micromorphological identification of faecalspherulites from archaeological and modern materials. Journal ofArchaeological Science 25, 435–444.

Canti, M. G. (1999). The production and preservation of faecalspherulites: animals, environment and taphonomy. Journal ofArchaeological Science 26, 251–258.

Courty, M. A., Goldberg, P. & Macphail, R. I. (1989). Soils andMicromorphology in Archaeology. Cambridge and New York:Cambridge University Press.

Courty, M. A., Macphail, R. I. & Wattez, J. (1991). Soil micromor-phological indicators of pastoralism; with special reference toArene Candide, Finale Ligure, Italy. Rivista di Studi Liguri, A.LVII, 127–150.

di Lernia, S. (1998). Cultural control over wild animals during theearly Holocene: the case of Barbary sheep in central Sahara. In (S.di Lernia & G. Manzi, Eds) Before Food Production in NorthAfrica. A.B.A.C.O. Edizioni, pp. 113–126.

Dyson-Hudson, R. & Dyson-Hudson, N. (1980). Nomadicpastoralism. Annual Review in Anthropology 15, 15–61.

Evershed, R. P., Bethell, P. H., Reynolds, P. J. & Walsh, N. J. (1997).5� stigmastanol and related 5�-stanols as biomarkers of manuring:analysis of modern experimental material and assessment of thearchaeological potential. Journal of Archaeological Science 24,485–495.

Galaty, J. G. (1981). Land and livestock among Kenyan Maasai.Journal of African Studies 16, 68–88.

Gebauer, A. B. & Price, T. D. (1992). Foragers to farmers: anintroduction. In (A. B. Gebauer & T. D. Price, Eds) Transitions toAgriculture in Prehistory. Chicago: Prehistory Press. Monographsin World Archaeology No. 4, pp. 1–10.

Gifford, D. P. (1978). Ethnoarchaeological observations of naturalprocesses affecting cultural materials. In (R. A. Gould, Ed.)Explanation in Ethnoarchaeology. Albuquerque: University ofNew Mexico Press, pp. 77–101.

Gifford-Gonzalez, D. P. (1998). Gender and early pastoralists inEast Africa. In (S. Kent, Ed.) Gender in African Prehistory. WalnutCreek: Altamira Press, pp. 115–137.

Goldberg, P. & Whitbread, I. (1993). Micromorphological study of aBedouin tent floor. In (P. Goldberg, D. T. Nash & M. D.Petraglia, Eds) Formation Processes in Archaeological Context.Madison, Wisconsin: Prehistory Press, pp. 165–188.

Goren, Y. (1999). On determining use of pastoral cave sites: a criticalassessment of spherulites in archaeology. Journal of The IsraelPrehistoric Society 29, 123–128.

Grandin, B. E., De Leeuw, P. N. & Lembuya, P. (1989). Drought,resource distribution, and mobility in two Maasai group ranches,southeastern Kajiado District. In (T. E. Downing, K. W. Gitu &C. M. Kamau, Eds) Coping with Drought in Kenya. Boulder andLondon: Lynne Rienner Publishers, pp. 245–263.

Harris, D. R. (1996). The origins and spread of agriculture andpastoralism in Eurasia: an overview. In (D. R. Harris, Ed.)The Origins and Spread of Agriculture and Pastoralism inEurasia. Washington, D.C.: Smithsonian Institution Press,pp. 552–573.

Jones, A. K. G., Hutchinson, A. R. & Nicholson, C. (1988). Theworms of Roman horses and other finds of intestinal parasite eggsfrom unpromising deposits. Antiquity 62, 275–276.

Geo-Ethnoarchaeology of Pastoral Sites 459

Karkanas, P., Bar-Yosef, O., Goldberg, P. & Weiner, S. (2000).Diagenesis in prehistoric caves: the use of minerals that form insitu to assess the completeness of the archaeological record.Journal of Archaeological Science 27, 915–929.

Kruuk, H. (1980). The effects of large carnivores on livestock andanimal husbandry in Marsabit District, Kenya. IPAL (IntegratedProject in Arid Lands) Technical Report E-4 (ITE Project 675),Unesco Programm on Man and the Biosphere.

Lamprey, R. & Waller, R. (1990). The Loita-Mara region inhistorical times: patterns of subsistence, settlement and eco-logical change. In (P. Robertshaw, Ed.) Early Pastoralists ofSouth-western Kenya. Nairobi: Memoirs of the British Institutein Eastern Africa no. 11, British Institute in Eastern Africa,pp. 16–35.

LeGeros, R. Z. (1991). Calcium Phosphates in Oral Biology andMedicine. In (H. M. Myers, Ed.) Monographs in Oral Science,vol. 15. Basel, Switzerland: Karger.

Lillios, K. T. (1992). Phosphate fractionation of soils at Agroal,Portugal. American Antiquity 57, 495–506.

Macphail, R. I. & Goldberg, P. (1995). Recent advances in micro-morphological interpretations of soils and sediments fromarchaeological sites. In (A. J. Barham & R. I. Macphail, Eds)Archaeological Sediments and Soils: Analysis, Interpretation andManagement. London: University College London, pp. 1–24e.

Marean, C. (1992). Hunter to herder: large mammal remains fromthe hunter-gatherer occupation at Enkapune ya Muto rockshelter,central rift, Kenya. African Archaeological Review 10, 65–127.

Marshall, F. (1990). Origins of specialized pastoral production inEast Africa. American Anthropologist 92, 873–894.

Marshall, F. (2000). The origins of domesticated animals in EasternAfrica. In (K. C. McDonald & R. M. Blench, Eds) The Origins andDevelopment of African Livestock: Archaeology, genetics, linguis-tics and ethnography. London: University College London Press,pp. 191–221.

Marshall, F. & Hildebrand, E. (2002). Cattle before crops:the beginnings of food production in Africa. Journal of WorldPrehistory, (in press).

Mbae, N. B. (1986). Aspects of Maasai Ethnoarchaeology:Implications for Archaeological Interpretation. Masters Thesis.University of Nairobi.

Middleton, W. D. & Price, T. D. (1996). Identification of activityareas by multi element characterization of sediments from modernand archaeological house floors using inductively coupled plasma-atomic emission spectroscopy. Journal of Archaeological Science23, 673–687.

Mulholland, S. C. & Rapp, G. Jr (1992). A morphological classifi-cation of grass silica bodies. In (G. Rapp, Jr & S. C. Mulholland,Eds) Phytolith Systematics. New York: Plenum Press, pp. 65–89.

Mutundu, K. K. (1999). Ethnohistoric Archaeology of the Mukogodoin north-central Kenya: Hunter-Gatherer Subsistence and theTransition to Pastoralism in Secondary Settings. BARInternational Series 775.

Powers, A. H., Padmore, J. & Gilbertson, D. D. (1989). Studies oflate prehistoric and modern opal phytoliths from coastal sand

dunes and machair in northwest Britain. Journal of ArchaeologicalScience 16, 27–45.

Price, T. D. & Gebauer, A. B. (1995). New perspectives on thetransition to agriculture. In (T. D. Price & A. B. Gebauer, Eds)Last Hunters, First Farmers. Santa Fe, New Mexico: School ofAmerican Research Press, pp. 3–19.

Robertshaw, P. T. (1978). The archaeology of an abandonedpastoralist camp-site. South African Journal of Science 74, 29–31.

Robertshaw, P. & Marshall, F. (1990). Ngamuriak. In (P.Robertshaw, Ed.) Early Pastoralists of South-western Kenya.Memoirs of the British Institute in Eastern Africa no. 11, BritishInstitute in Eastern Africa, Nairobi, pp. 54–72.

Ryan, K., Munene, K., Kahinju, S. M. & Kunoni, P. K. (2000).Ethnographic perspectives on cattle management in semi-aridenvironments: a case study from Maasailand. In (K. C. McDonald& R. M. Blench, Eds) The Origins and Development of AfricanLivestock: Archaeology, genetics, linguistics and ethnography.London: University College London Press, pp. 462–477.

Sanchez, A., Canabate, M. L. & Lizcano, R. (1996). Phosphorusanalysis at archaeological sites: an optimization of the method andinterpretation of the results. Archaeometry 38, 151–164.

Schelvis, J. (1992). The identification of archaeological dung depositson the basis of remains of predatory mites (Acari; Gamasida).Journal of Archaeological Science 19, 677–682.

Simpson, I. A., Dockrill, S. J., Bull, I. D. & Evershed, R. P. (1998).Early anthropogenic soil formation at Tofts Ness, Sanday,Orkney. Journal of Archaeological Science 25, 729–746.

Smith, A. B. (1992). Pastoralism in Africa: Origins and DevelopmentEcology. Ohio University Press.

Thomas, G. W. (1996). Soil pH and soil acidity. In Methods of SoilAnalysis. Part 3. chemical Methods. Soil Science Society ofAmerica, Book Series no. 5, pp. 475–490.

Touber, L. (1983). Soils and vegetation of the Amboseli-Kibweziarea. In (B. J. A. van der Pouw & V. W. P. van Engelen, Eds)Reconnaissance Soil Survey Report No. R6. Nairobi: Kenya SoilSurvey.

van der Veen, M. (1999). The Exploitation of Plant Resources inAncient Africa. New York: Klewer Academic/Plenum Publishers.

Wattez, J., Courty, M. A. & Macphail, R. I. (1990). Burnt organo-mineral deposits related to animal and human activities inprehistoric caves. In (L. A. Douglas, Ed.) Soil Micromorphology:A Basic and Applied Science. Amsterdam: Elsevier, pp. 431–439.

Weiner, S., Goldberg, P. & Bar-Yosef, O. (1993). Bone preservationin Kebara Cave, Israel using on-site Fourier Transform InfraredSpectrometry. Journal of Archaeological Science 20, 613–627.

Western, D. & Dunne, T. (1979). Environmental aspects of settle-ment site decisions among pastoral Maasai. Human Ecology 7,75–98.

Yellen, J. E. (1984). The integration of herding into prehistorichunting and gathering economics. In (A. Hall, G. Avery, D. M.Avery, M. L. Wilson & J. B. Humphreys, Eds) Frontiers: SouthernAfrican Archaeology Today. Cambridge Monographs in AfricanArchaeology 10, BAR International Series 207, pp. 53–64.